This study systematically evaluates the thermal and humidity control performance of polymer-dispersed liquid crystal (PDLC) smart windows in an operational subtropical commercial building. Conducted from September to November 2025 at the China Railway Construction Building in Zhuhai, China, the field experiment compared four configurations: conventional curtains (fully deployed and fully retracted, respectively) and PDLC film in transparent and opaque states. Results demonstrate that during the high-solar-radiation period (September–October), PDLC in the opaque state exhibited superior thermal control, limiting interior temperature increases to only 2% of the magnitude observed in the transparent state and yielding a maximum interior surface temperature difference of 1.88 °C during peak solar hours (14:00 to 17:00). Humidity fluctuations remained exceptionally stable at ±1.5% in frosted state, significantly outperforming traditional curtain systems (±5.1% to ±8.9%). During November’s transitional climate, the frosted state continued providing thermal buffering, reducing indoor temperature rise by approximately 0.37 °C compared to the transparent state, while the transparent configuration maintained relative humidity approximately 0.5% higher—potentially beneficial for mitigating winter dryness. Cross-seasonal analysis revealed a 57% reduction in indoor temperature rise (from 3.06 °C to 1.31 °C) between September–October and November, directly attributable to seasonal variations in solar geometry. These findings confirm PDLC smart windows’ ability to dynamically regulate temperature, humidity, and daylighting across different seasonal conditions. Despite limitations including non-uniform room geometries and single-climate validation, this research establishes PDLC technology as a promising solution for energy-efficient building envelopes in subtropical regions. Future work should focus on standardized comparative testing, multi-climate validation, long-term durability assessment, and integration with building automation systems.
Sun et al. (Wed,) studied this question.